Open-SWE-Traces: Advancing Dual-Mode Multilingual Distillation for Software Engineering Agents
arXiv:2606.16038v1 Announce Type: cross Abstract: The path toward autonomous software engineering is currently bottlenecked by a severe deficit of diverse, large-scale trajectory data. We address this by introducing \ourdataset, an expansive dataset of 207,489 agentic trajectories spanning nine programming languages (Python, Go, TS, JS, Rust, Java, PHP, C, C++). Sourced from 20,000 real-world PRs via OpenHands and SWE-agent harnesses, the dataset utilizes a hybrid-reasoning synthesis: Minimax-M2.5 generates trajectories with explicit "thinking" processes, while Qwen3.5-122B provides high-quali
The rapid advancement of large language models and the push for autonomous systems necessitate more robust and diverse training data for software engineering agents.
This dataset directly addresses a critical bottleneck in the development of capable software engineering AI agents, accelerating their potential for autonomy across various programming environments.
The availability of a large, diverse dataset of agentic trajectories for software development will significantly improve the training and performance of AI agents in software engineering.
- · AI Agent Developers
- · Software Development Companies
- · Open-source AI Community
- · Cloud Computing Providers
- · Monolithic Software Development Teams
- · Manual Code Reviewers
Improved performance and broader applicability of AI software engineering agents.
Reduced software development cycles and increased automation in coding and debugging tasks.
Potential for AI agents to independently develop and maintain complex software systems with minimal human oversight.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.AI